|
Activity Number:
|
271
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #309407 |
|
Title:
|
Analyzing Input and Structural Uncertainty of a Hydrological Model with Stochastic Time-Dependent Parameters
|
|
Author(s):
|
Peter Reichert*+ and Johanna Mieleitner
|
|
Companies:
|
Swiss Federal Institute of Aquatic Science and Technology and Eawag
|
|
Address:
|
Eawag, Duebendorf, 8600, Switzerland
|
|
Keywords:
|
model structure deficit ; time-dependent parameters ; bias
|
|
Abstract:
|
A recently developed technique for analyzing the causes of bias in results of deterministic models is applied to a simple hydrological model. The application of this technique to a simple 7-parameter conceptual hydrological model demonstrates the power of the technique. The parameters (including additional parameters for input modification) have significantly different potential for bias reduction. Although there is a minor potential for improvement of the underlying deterministic model, most of the bias seems to be caused by random effects, particularly rainfall input uncertainty. Propagation of these errors through the model leads to strongly correlated model output errors because of the internal storage mechanisms in the model.
|
